Bioinformatics tools in precision medicine

The radical technological advances of sequencing techniques that followed the completion of the Human Genome Project in the beginning of this century have given rise to an unprecedented volume and variety of -omics data, that, combined with phenotypic information have supported the evolution of genomic medicine. This dissertation was set to explore the potential of bioinformatics to further enable personalized medicine research with a focus on pharmacogenomics, by designing and implementing solutions aspiring to facilitate the translation of pharmacogenomic knowledge into actions.

Towards this end, publicly available genomic data resources were exploited to describe the spectrum, complexity and structure of pharmacogenomic variation. The exploratory analysis the DiscovEHR cohort highlighted the abundance of rare single-nucleotide variants and the need for functional assessment of novel variation in genes implicated in drug response. In addition, a comprehensive map of the ethnogeographic variability of variants underlying G6PD deficiency was constructed, emphasizing differences in allele frequencies and composition across various populations. Moreover, FINDbase, a database documenting the population frequencies of disease- and drug response-elated variants was refined, both in terms of architecture, performance and data contents.

As a further step, supervised machine learning approaches were explored for the extraction of pharmacogenomic associations from the biomedical literature, as well as for the computational functional assessment of novel pharmacovariants. Finally, a generic model, optimized for the analysis of clinical trial data was developed, providing a user-friendly web interface to facilitate the comprehensive cost-utility assessment of genome-guided interventions.